Cascade Control of Antagonistic VSA—An Engineering Control Approach to a Bioinspired Robot Actuator
A cascade control structure for the simultaneous position and stiffness control of antagonistic tendon-driven variable stiffness actuators (VSAs) implemented in a laboratory setup is presented in the paper. Cascade control has the ability to accelerate, additionally stabilize, and reduce oscillation...
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Frontiers Media S.A.
2019-09-01
|
Series: | Frontiers in Neurorobotics |
Subjects: | |
Online Access: | https://www.frontiersin.org/article/10.3389/fnbot.2019.00069/full |
id |
doaj-450453ec77f14ca1911356a11d597f3f |
---|---|
record_format |
Article |
spelling |
doaj-450453ec77f14ca1911356a11d597f3f2020-11-24T20:47:58ZengFrontiers Media S.A.Frontiers in Neurorobotics1662-52182019-09-011310.3389/fnbot.2019.00069468971Cascade Control of Antagonistic VSA—An Engineering Control Approach to a Bioinspired Robot ActuatorBranko LukićKosta JovanovićTomislav B. ŠekaraA cascade control structure for the simultaneous position and stiffness control of antagonistic tendon-driven variable stiffness actuators (VSAs) implemented in a laboratory setup is presented in the paper. Cascade control has the ability to accelerate, additionally stabilize, and reduce oscillations, which are all extremely important in systems such as a tendon-driven compliant actuators with elastic transmission. Inner-loop controllers are closed in terms of motor positions, and outer-loop controllers in terms of actuator position and estimated stiffness. The dominant dynamics of the system (position and stiffness), composed of the mechanical part and inner loops, are identified by a closed-loop auto-regressive with exogenous input (ARX) model. The outer-loop controllers are tuned on the basis of experimentally identified transfer functions of the system in several nominal operating points for different stiffness values. After the system is identified, a controller bank is generated in which a pair of actuator position and stiffness controllers correspond to a nominal operating point and covers the area surrounding the nominal point for which it is designed. The controllers used are integral-proportional differential (I-PD) and integral-proportional (I-P) controllers, which are a variation of the PID and PI controllers with dislocated proportional and derivative gains from a direct to feedback branch that result to no overshoot for even fast reference changes (i.e., step signal), which is essential for preventing tendon slackening (meeting the pulling constraint). Analytical formulas for controller tuning based on only one parameter, λ, are also presented. Since position and stiffness loops are decoupled, it is possible to change λ for both loops independently and adjust their performance separately according to the needs. Also, the controller structure secures the smooth response without overshooting step reference or step disturbance signal, which make practical implementation possible. After all the controllers were designed, the cascade control structure for simultaneous position and stiffness control was successfully evaluated in a laboratory setup. Thus, the presented control approach is simple to implement, but with a performance that ensures a pulling constraint for tendon-driven actuators as a foundation for bioinspired antagonistic VSAs.https://www.frontiersin.org/article/10.3389/fnbot.2019.00069/fullantagonistic actuatortendon-driven actuatorsvariable stiffness actuatorsbioinspired roboticsphysical human–robot interactionposition–stiffness control |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Branko Lukić Kosta Jovanović Tomislav B. Šekara |
spellingShingle |
Branko Lukić Kosta Jovanović Tomislav B. Šekara Cascade Control of Antagonistic VSA—An Engineering Control Approach to a Bioinspired Robot Actuator Frontiers in Neurorobotics antagonistic actuator tendon-driven actuators variable stiffness actuators bioinspired robotics physical human–robot interaction position–stiffness control |
author_facet |
Branko Lukić Kosta Jovanović Tomislav B. Šekara |
author_sort |
Branko Lukić |
title |
Cascade Control of Antagonistic VSA—An Engineering Control Approach to a Bioinspired Robot Actuator |
title_short |
Cascade Control of Antagonistic VSA—An Engineering Control Approach to a Bioinspired Robot Actuator |
title_full |
Cascade Control of Antagonistic VSA—An Engineering Control Approach to a Bioinspired Robot Actuator |
title_fullStr |
Cascade Control of Antagonistic VSA—An Engineering Control Approach to a Bioinspired Robot Actuator |
title_full_unstemmed |
Cascade Control of Antagonistic VSA—An Engineering Control Approach to a Bioinspired Robot Actuator |
title_sort |
cascade control of antagonistic vsa—an engineering control approach to a bioinspired robot actuator |
publisher |
Frontiers Media S.A. |
series |
Frontiers in Neurorobotics |
issn |
1662-5218 |
publishDate |
2019-09-01 |
description |
A cascade control structure for the simultaneous position and stiffness control of antagonistic tendon-driven variable stiffness actuators (VSAs) implemented in a laboratory setup is presented in the paper. Cascade control has the ability to accelerate, additionally stabilize, and reduce oscillations, which are all extremely important in systems such as a tendon-driven compliant actuators with elastic transmission. Inner-loop controllers are closed in terms of motor positions, and outer-loop controllers in terms of actuator position and estimated stiffness. The dominant dynamics of the system (position and stiffness), composed of the mechanical part and inner loops, are identified by a closed-loop auto-regressive with exogenous input (ARX) model. The outer-loop controllers are tuned on the basis of experimentally identified transfer functions of the system in several nominal operating points for different stiffness values. After the system is identified, a controller bank is generated in which a pair of actuator position and stiffness controllers correspond to a nominal operating point and covers the area surrounding the nominal point for which it is designed. The controllers used are integral-proportional differential (I-PD) and integral-proportional (I-P) controllers, which are a variation of the PID and PI controllers with dislocated proportional and derivative gains from a direct to feedback branch that result to no overshoot for even fast reference changes (i.e., step signal), which is essential for preventing tendon slackening (meeting the pulling constraint). Analytical formulas for controller tuning based on only one parameter, λ, are also presented. Since position and stiffness loops are decoupled, it is possible to change λ for both loops independently and adjust their performance separately according to the needs. Also, the controller structure secures the smooth response without overshooting step reference or step disturbance signal, which make practical implementation possible. After all the controllers were designed, the cascade control structure for simultaneous position and stiffness control was successfully evaluated in a laboratory setup. Thus, the presented control approach is simple to implement, but with a performance that ensures a pulling constraint for tendon-driven actuators as a foundation for bioinspired antagonistic VSAs. |
topic |
antagonistic actuator tendon-driven actuators variable stiffness actuators bioinspired robotics physical human–robot interaction position–stiffness control |
url |
https://www.frontiersin.org/article/10.3389/fnbot.2019.00069/full |
work_keys_str_mv |
AT brankolukic cascadecontrolofantagonisticvsaanengineeringcontrolapproachtoabioinspiredrobotactuator AT kostajovanovic cascadecontrolofantagonisticvsaanengineeringcontrolapproachtoabioinspiredrobotactuator AT tomislavbsekara cascadecontrolofantagonisticvsaanengineeringcontrolapproachtoabioinspiredrobotactuator |
_version_ |
1716809441339244544 |